Skip to content
B2BenrichA ContactLevel company

Lead scoring with intent data: fit, tech stack and timing in one score

A simple scoring model a GTM engineer can build: firmographic fit, tech-stack fit and intent timing, weighted into one number per account.

Dag Holmen · Updated · 5 min read

View as Markdown

Most lead scoring models only see what people do with you: form fills, email clicks, page views. That misses the companies researching your category somewhere else, and it says little about timing. Intent data lead scoring fixes the second part. You keep the fit criteria you already trust, add tech-stack fit, and add a timing score based on what each account is researching right now.

This guide builds that model as code: three parts, weighted into one number per account, then cut into tiers. The weights and thresholds below are a starting point. You’ll tune them against your own closed deals.

The three parts of the score

PartQuestion it answersData sourceStarting weight
FitCould this company buy from us?Company enrichment and search40 points
Tech stackDoes their stack make us a better fit?Prospector technologies filter25 points
TimingAre they researching our topics now?Intent data35 points

Fit and tech change slowly, so you can compute them weekly or monthly. Timing changes daily, so it gets recomputed on every run. Keeping them as separate sub-scores also makes the final number easy to explain. A rep can see that an account scored high on timing but only medium on fit.

Fit score from firmographics

Pick five or so criteria that separate your won deals from your lost ones. Company records from POST /v1/enrich/companies/domain-to-company or POST /v1/search/companies give you the usual inputs: industry, employee range, revenue range and headquarters country. A simple rule set is enough to start:

  • Industry in your core list: 15 points
  • Employee range in your sweet spot: 15 points
  • Headquarters in a region you sell to: 10 points

Tech-stack score

Technographics answer a question firmographics can’t: will this company actually work with your product? If you integrate with HubSpot, a HubSpot shop is a better fit than a company of the same size on something else. Technographic data covers the categories in more depth.

The prospector takes a technologies company filter, so you can build a set of domains per tool and check each account against it.

const BASE = 'https://api.b2benrich.com';
const headers = { 'X-Api-Key': process.env.B2BENRICH_API_KEY, 'Content-Type': 'application/json' };

async function domainsUsing(technology, companyFilters = {}) {
  const domains = new Set();
  let page = 1;
  while (true) {
    const res = await fetch(`${BASE}/v1/search/prospector`, {
      method: 'POST',
      headers,
      body: JSON.stringify({
        contact_filters: {},
        company_filters: { ...companyFilters, technologies: technology },
        page,
        page_size: 100
      })
    }).then((r) => r.json());
    res.records.forEach((p) => domains.add(p.company_domain));
    if (!res.has_more) break;
    page = res.next_page;
  }
  return domains;
}

const hubspotDomains = await domainsUsing('HubSpot', { hq_country: 'United States' });

Get the exact technology names from POST /v1/search/meta/technologies. Adding your ICP filters to company_filters keeps each set small.

Timing score from intent

The timing score turns raw signals into one number. Three things matter.

  1. Topic type. Commercial topics (pricing, alternatives, competitor names) count for more than informational ones.
  2. Recency. A signal from yesterday counts for more than one from three weeks ago. A half-life of seven days is a reasonable default.
  3. Breadth. Research on several related topics, or by more than one person, points to a real project.

Put it together

// TODO: replace with the B2Benrich intent endpoint once it's documented.
// Returns your own shape: [{ domain, topic, strength (0-1), seenAt }]
async function getIntentSignals(accountDomains, topics) {
  return [];
}

const COMMERCIAL = new Set(['competitor-x alternatives', 'sales engagement pricing']);
const HALF_LIFE_DAYS = 7;

function fitScore(company) {
  let s = 0;
  if (CORE_INDUSTRIES.includes(company.industry)) s += 15;
  if (SWEET_SPOT_RANGES.includes(company.employees_range)) s += 15;
  if (SALES_REGIONS.includes(company.hq_country)) s += 10;
  return s; // 0-40
}

function techScore(domain) {
  let s = 0;
  if (hubspotDomains.has(domain)) s += 15;   // a tool you integrate with
  if (competitorDomains.has(domain)) s += 10; // a tool you replace
  return s; // 0-25
}

function timingScore(signals, now = Date.now()) {
  let s = 0;
  for (const sig of signals) {
    const ageDays = (now - new Date(sig.seenAt)) / 86_400_000;
    const decay = Math.pow(0.5, ageDays / HALF_LIFE_DAYS);
    const weight = COMMERCIAL.has(sig.topic) ? 1 : 0.5;
    s += 20 * sig.strength * weight * decay;
  }
  const topics = new Set(signals.map((x) => x.topic));
  if (topics.size >= 2) s += 5;
  return Math.min(Math.round(s), 35); // 0-35
}

function score(company, signals) {
  const fit = fitScore(company);
  const tech = techScore(company.domain);
  const timing = timingScore(signals);
  return { fit, tech, timing, total: fit + tech + timing };
}

Store all four numbers, not just the total. When the score looks wrong, the parts tell you which input to fix.

Turn the score into tiers

TierTotal scoreWhat happens
A70 and upRoute to a rep today
B45 to 69Sequence plus ads, re-score next run
CBelow 45Marketing only

Add one rule on top: an account with a timing score of zero never reaches tier A, however good the fit. That keeps the A tier about timing, which is the point of adding intent. Routing tier A accounts to a sequence the day they cross the line is covered in signal-based outbound.

Back-test, then tune

Run the model on accounts from your last 12 months of deals and check where the wins landed. If closed-won accounts sit mostly in tier B, your thresholds are too high or a weight is off. Change one thing at a time and re-run.

Once you have a few hundred scored accounts with outcomes, you can try predictive lead scoring: fit a logistic regression or a small tree model on the same features and compare it to the rules. Keep the rules if the model doesn’t clearly beat them. A score a rep can explain is easier to trust.

Re-score on a schedule. Nightly works well, because timing decays every day. B2Benrich is unlimited API usage for $500 a month, no credits, so a nightly re-score across your full in-market account list costs the same as a weekly one.

Frequently asked questions

What is intent data lead scoring?
Intent data lead scoring adds a timing component to a lead or account score, based on what the company is researching right now, next to the usual fit criteria like industry, size and tech stack.
How do I weight fit, tech stack and intent?
A reasonable starting point is 40 points for fit, 25 for tech stack and 35 for intent timing. Treat those as a first guess and tune them by checking where your past closed-won accounts would have scored.
How should intent signals decay in a lead scoring model?
Give each signal a half-life, for example seven days, so its weight halves every week. That keeps old research from holding an account at the top of the list.
Is predictive lead scoring better than a rules-based model?
It can be once you have enough scored accounts with known outcomes. Start with rules you can explain, then test a model on the same features and switch only if it clearly ranks your wins higher.
Can I use technographics in lead scoring?
Yes. Award points when an account uses a tool you integrate with or a tool you replace. With B2Benrich you can build those sets with the prospector’s technologies filter.

Get your API key.

Sign up, make your first call and start building your lists.

We’ll email you a link to log in. · Read the docs